Tunable giant negative thermal expansion in Ti2O3-based polycrystalline materials

Thermal expansion properties are investigated for Ti2O3-based sintered polycrystalline materials by controlling crystal lattice parameters and porous structures. Negative thermal expansion (NTE) is observed in a temperature range between room temperature and 593 K, and found to exceed ΔL/L = −0.7% at high temperatures, in spite of continuous increase of the unit cell volume. The temperature range for the significant NTE is successfully tuned by modifying electron-lattice-coupled phase-transition temperature. The magnitude of the NTE is enhanced upon introducing pores, indicating the importance of microstructural effect. These results demonstrate high potential of Ti2O3-based materials for practical applications.

T hermal expansion is an important ingredient to be considered in various kinds of industrial domains. Especially at the interface between different materials, bending, cracking and peeling due to difference in coefficient of thermal expansion of the adjacent materials are serious problems in many applications, therefore it is necessary to control thermal expansion properties of the materials. 1,2) Negative thermal expansion (NTE) materials, which contract in volume as temperature is increased, have recently attracted great attention. 1,[3][4][5] Due to intensive researches, many kinds of NTE materials have been discovered, for example, β-eucryptite, 6) ZrW 2 O 8 , 7) antiperovskite manganese nitrides, 8,9) LaCu 3 Fe 4 O 12 10) BiNiO 3 -based materials, [11][12][13] V 2 OPO 4 , 14,15) Ca 2 RuO 4 -based materials, 16,17) and Cu 1.8 Zn 0.2 V 2 O 7 . 18,19) From the view point of practical use of NTE materials, large values of NTE are desirable. Conventionally, bulk NTE due to unit cell volume contraction upon increasing temperature was discussed in ZrW 2 O 8 7) and manganese nitrides. 8,9) Recently, enhanced NTE arising from microstructural effects was found in layered ruthenate Ca 2 RuO 4 and its derivatives. 16,17) The observed magnitude of NTE is much larger than its contraction of the unit cell volume, therefore it is also important to consider the microstructural effects due to crystalline grains, which collectively form pores and individually exhibit anisotropic expansion. 16,17) Such a microstructural origin of NTE has been also reported for Cu 1.8 Zn 0.2 V 2 O 7 , 18,19) and exploiting the microstructural effect is one of the most promising strategies to obtain large magnitude of NTE. 1) In addition to the magnitude of NTE value, controlling the temperature range of NTE is also important from the perspective of industrial applications for a wide range of usages. Furthermore, in terms of an environmental aspect, NTE materials should be desirably composed of abundant elements without toxicity.
Recently, giant NTE in a wide temperature range between room temperature and 600 K was reported for sintered Ti 2 O 3 polycrystalline sample, 20) but the mechanism of NTE is still unclear. Ti 2 O 3 crystallizes in a corundum structure, whose schematic structure is shown in the inset to Fig. 1(a). Thermally induced metal-insulator transition around 473 K was reported for Ti 2 O 3 and the phase transition is considered to be caused by the destruction of Ti 3+ -Ti 3+ singlet dimer upon heating, which results in the elongation (contraction) of c-axis (a-axis) lattice constant. 21,22) Anisotropic deformation of unit cell of Ti 2 O 3 through the phase transition was reported previously, 23) and may be related to the NTE observed in the polycrystalline Ti 2 O 3 samples.
In this study, we investigated the NTE in polycrystalline Ti 2 O 3 -based materials by controlling lattice parameters and porous structures. The temperature range where the NTE is observed is successfully tuned by controlling the phase transition temperature via chemical modification and introducing oxygen deficiency. Although the unit cell volume determined by powder X-ray diffraction measurements increases by approximately 0.5% as temperature is raised, the macroscopic NTE is observed in a wide temperature range between room temperature and 593 K, and found to exceed ΔL/L = −0.7% at high temperatures. We also find that the magnitude of NTE is affected by the total amount of pore volume in sintered polycrystalline samples. This indicates that the microstructural effect plays an important role in the NTE of this material.
All polycrystalline samples were synthesized by a spark plasma sintering (SPS) process. Starting materials [Ti 2 O 3 : 99.9% (averaged particle size: 41 μm and 12 μm denoted as large and small particle, respectively), Ti: 99.95%, Nb 2 O 3 : 99.9% in purity] with a prescribed ratio were mixed and pressed into pellets, and then subject to the SPS process in an argon atmosphere under a pressure of 40 MPa for 10 min at respective temperatures. In this work, 7 samples were investigated in total. The obtained samples were checked by using a powder X-ray diffractometer (Smart Lab, Rigaku), and confirmed that main phase was of corundum-type Ti 2 O 3 at ambient temperature. Cumulative pore volume was obtained by mercury intrusion porosimetry using Micromeritics AutoPore IV 9520. Table I summarizes starting materials, sintering temperature, lattice constants at ambient temperature and porosities of sample A-D and G. We use nominal composition of each sample throughout this paper. Hightemperature X-ray powder diffraction measurements were performed for 303-673 K in Ar-flow condition. Linear thermal expansion ΔL/L was measured for 303-593 K using thermo-mechanical property analysis equipment (Thermo plus EVO2 8310, Rigaku). Resistivity was measured in a helium atmosphere from 300 up to 593 K by using a commercial apparatus (ZEM-3, ADVANCE-RIKO, Inc.). Figure 1(a) shows temperature dependence of resistivity for selected samples. The metal-insulator transition is observed around 450-500 K as a steep increase in resistivity upon cooling in samples (D) and (E), while it takes place in Nb-doped sample (G) at lower temperatures as compared with the samples (D) and (E), indicating good tunability of the transition temperature in terms of Nb-doping. Figure 1(b) shows the T-dependence of the linear thermal expansion ΔL/L (303 K) ≡ (L(T) − L(303 K))/L(303 K) for all samples. NTE is observed in a wide temperature range between room temperature and 593 K, and the magnitude (|ΔL/L|) exceeds 0.7% for sample (D) around 590 K. Figure 1(c) displays the coefficient of linear thermal expansion (α) derived from the slope between ΔL(T − 10 K)/L and ΔL(T + 10 K)/L. Nb-doping effectively lowers the temperature region where the magnitude of α takes a maximum, as evident in sample (G). Oxygen deficiency leads to the increase in the temperature range of NTE, as discerned in sample (E).
To unveil microscopic mechanism of the NTE in Ti 2 O 3 -based polycrystalline materials, we conducted hightemperature powder X-ray diffraction analysis to deduce the temperature-dependent lattice constants. Here, to clarify how the anisotropic deformation of a unit cell contributes to the bulk NTE, we introduce an anisotropy parameter R defined as R = a/c, following Ref. 18. Lattice constants, anisotropy parameter R, and unit cell volume are plotted in Figs. 2(a)-2(d) as a function of temperature for selected samples. Clearly, structural evolution associated with the destruction of Ti 3+ -Ti 3+ singlet dimer is observed as elongation of c-axis and contraction of a-axis as the temperature is increased. In the Nb-doped samples (F) and (G), the temperature range of the rapid structural evolution is lowered. On the other hand, in the oxygen-deficient sample (E), the temperature range of the phase transition is shifted to higher side. Figure 2(d) clearly shows that unit cell volume for each sample increases as temperature is elevated, and in case of sample (B), for example, volume expansion by approximately 0.5% is observed between room temperature and 573 K. This indicates that the observed bulk NTE cannot be attributed to the feature of a unit cell volume and that microstructural effects should be taken into account.
In Fig. 3, we replot the bulk volume expansion estimated from the thermo-mechanical measurements (3ΔL/L) against the change in anisotropy parameter ΔR/R. A linear relationship between the bulk volume expansion and the change in anisotropy parameter is obvious, indicating that the anisotropic deformation of unit cell in the polycrystalline material is responsible for the macroscopic NTE. For Nb-doped samples (F) and (G), the relationship slightly deviates from the linear one. This deviation may be caused by the existence of the certain amount of impurity phase in the Nb-doped samples.
To investigate the possible role of porous structure quantitatively, we obtained cumulative pore volume for several samples by the mercury intrusion method, and the results are presented in Fig. 4(a). For each sample, distribution of the pore radius from sub-micrometer to micrometer scale is observed as rapid change in the cumulative pore volume. By using the cumulative pore volume, we can calculate the total porosity in each sample.   Table I. Starting materials, sintering temperature, and lattice constants (a and c) of the hexagonal unit cell at ambient temperature. In starting material, Ti 2 O 3 (L) and (S) denote Ti 2 O 3 powder of large size particle (averaged particle size: 41 μm) and small size particle (averaged particle size: 12 μm), respectively. The porosity of each sample measured by mercury intrusion porosimetry is shown for sample A-D and G.

No
Nominal  Figure 4(b) shows the relation between the total porosity and the bulk volume expansion (3ΔL/L) as estimated by the thermo-mechanical measurement. As the total porosity increases, the magnitude of the NTE between 303 and 593 K is enhanced. This relationship indicates that the porous structure effectively contributes to the observed bulk NTE through the anisotropic deformation of each grain. Table II shows the relevant parameters for reported NTE materials. In some materials, bulk NTE values as obtained by dilatometry or thermo-mechanical measurements are greater than those deduced from the crystallographic unit cell volume. For example, the bulk NTE value was enhanced over that of an unit cell in CaRuO 4 -based sintered materials and it was ascribed to the microstructural effect arising from the polycrystalline grains with large anisotropic deformation. 16,17) In comparison with other NTE materials where the microstructural effects are important, the unique feature to the Ti 2 O 3 -based materials in this work is that NTE is observed in bulk poly-crystals in spite of the positive thermal expansion of a unit cell. This significant difference in the thermal expansion property between bulk and a crystallographic unit cell highlights the extremely important role of the microstructural effect in this material.
The major accomplishment of the present work in comparison with the previous studies 1, [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] is the demonstration of the tunability of the temperature range where the NTE emerges. As shown in Figs. 1(a)-1(c), pristine Ti 2 O 3 of sample (D) has the largest absolute value of NTE coefficient around approximately 473 K that corresponds to the metalinsulator transition temperature, indicating that the electronic state is strongly coupled with crystal lattice in this system. Presumably, doped Nb prevents from Ti 3+ -Ti 3+ dimer formation and therefore the metal-insulator transition temperature shifts to lower temperature region, and as a result, the temperature range of NTE also shifts to a lower one due to the strong electron-lattice coupling. On the other hand, oxygen deficiency increases the phase transition temperature. Reduction in the number of hole-type carriers by the oxygen deficiencies is probably responsible for the observed shift of phase transition temperature to higher temperatures.
Another important achievement is the quantitative analysis for the effects of porous structure and anisotropic deformation of unit cell on the NTE. In the conventional NTE materials, the macroscopic NTE due to contraction of a unit cell is observed upon heating 7,8) and recently the NTE effect has been found to be enhanced by the microstructural effects in strongly correlated materials, [16][17][18][19] but the link between the NTE and the effects of porous structure and anisotropic deformation has remained elusive. The linear relationship between anisotropy parameter and the NTE (Fig. 3) indicates     that the anisotropic deformation of respective crystalline grains is relevant to the bulk NTE. Also, the magnitude of bulk NTE is found to be enhanced by increasing the porosity [ Fig. 4(b)]: cumulative pore volume measurement shows that the total porosity in bulk polycrystalline material is well correlated with the NTE, indicating that microstructural effects are important for the NTE.
In summary, we have investigated the NTE for Ti 2 O 3 -based polycrystalline materials by changing lattice parameters and porous structures. The temperature range of NTE is demonstrated to be successfully tuned by controlling the phase transition temperature via chemical doping. Although the unit cell volume monotonically increases as temperature is elevated, the bulk NTE is observed in a wide temperature range between room temperature and 593 K, and exceeds |ΔL/L| = 0.7% at high temperatures. The NTE is correlated with the anisotropic deformation of a unit cell, and the magnitude is enhanced as total pore volume in sintered polycrystalline samples is increased. These observations indicate that the microstructural effects play an important role in the NTE of this material. In view of the versatile tunability of the temperature range for NTE as well as environment-related aspects, Ti 2 O 3 -based materials possess high potential for practical applications.